Lately, it has become popular to assert that antidepressants don't work, for example see this newspaper article.
It is true that in many antidepressant trials the observed effects have been modest, and the "placebo response" rates have been very high. However, the recovery rate from depression in people who are untreated is not always indicative of a placebo response. A placebo response implies that the recovery was due to the non-specific benefits of treatment (hope etc.).
The high rate of spontaneous recovery reflects the episodic nature of clinical depression. Many episodes of depression are brief (ie. they recover spontaneously and do not need treatment), but as they become more chronic the recovery rate diminishes. Many clinical trials actively recruit people with mild episodes and no comorbidity - exactly the group that is likely to recover rapidly.
So, despite the popularity of asserting that antidepressants don't work for depression, the reality is simpler: not all episodes of depression (especially mild ones) benefit from antidepressants.